A watershed-scale evapotranspiration model considering forest type, stand parameters, and climate factors

We developed a model to estimate evapotranspiration (ET) for temperate coniferous and broadleaf forests that considers stand structure (density and height) and meteorological data (precipitation, air temperature, solar radiation, and vapor pressure deficit) as input. We used globally available data from 65 publications, as well as 37 datasets from sources such as FLUXNET, that were collected mainly from temperate climates. The wet canopy evaporation (Ewet) submodel consisted of a nonlinear function to determine the ratio of interception loss to precipitation from stand density for coniferous forests or stand height for broadleaf forests. The parameters of the Ewet submodel were determined separately for rainfall and snowfall. The dry canopy evaporation (Edry) submodel consisted of a simplified Penman-Monteith equation, including a nonlinear function to determine the reference surface conductance from stand height. We tested the prediction ability of the model using data from three locations covered by evergreen coniferous (ET, 13-83 mm/month) and deciduous coniferous forests (24-104 mm/month) in central Japan, and a deciduous broadleaf forest (1-59 mm/month) in northern Wisconsin, USA, for a mixture of rain and snow. Our model reproduced seasonal ET patterns, with mean absolute errors of 10 +/- 5, 12 +/- 7, and 6 +/- 9 mm/month for the evergreen coniferous, deciduous coniferous, and deciduous broadleaf forests, respectively. The model is applicable to ET estimation over watershed-scale areas with heterogeneous biotic (forest condition) and abiotic (meteorological) factors in landscapes with complex topography.